en.Wedoany.com Reported - Physical AI and embodied intelligence company Kuawei Intelligence has completed a Series B funding round of 1 billion yuan, with a post-investment valuation exceeding 10 billion yuan, and plans to initiate an initial public offering. Investors participating in this round include Shenzhen Capital Group, Guiyang Digital Economy Fund, Qianhai Mother Fund, Lens Technology, ICBC Capital, Hengjian Asset, and Zhurui Capital, among others. Among them, Shenzhen Capital Group and Guiyang Digital Economy Fund have made consecutive heavy investments in two rounds; existing shareholders Nanshan Zhanxin Investment, Chengdu Science and Technology Innovation Investment, and Sichuan Academician Fund continued to follow the investment. The funds from this round will be primarily used for iterative development of underlying world model algorithms, upgrades to the physics engine, data infrastructure construction, enhancement of humanoid robot capabilities, and real-world scenario deployment.
Since its establishment in 2021, Kuawei Intelligence has focused on full-stack self-development of physical AI and world models, demonstrating a certain degree of foresight in its technical route selection. The progress of its layout in multiple core nodes has shown a faster pace compared to overseas industry giants such as NVIDIA and DeepMind. The company's embodied world model, DexWorldModel, recently topped the global authoritative embodied world model benchmark, WorldArena. This model is built around the real operational laws of the three-dimensional physical world, possessing latent space dynamic modeling capabilities. It enables unified coordinate system alignment and precise matching of spatiotemporal representations, addressing issues such as inaccurate spatial perception and lack of physical modeling in traditional vision solutions, supporting robots in completing practical tasks like sorting, assembly, and dual-arm composite operations.
In terms of data infrastructure, Kuawei Intelligence has innovatively proposed the Dexterity-BEV solution, which aligns visual input, robot body state, and action targets into a unified spatial reference frame. This establishes a standardized training foundation for multi-source heterogeneous robot data, solving problems such as scattered data standards, difficulty in transferring action skills, and challenges in reusing operational experience across different robot bodies. Relying on its four-layer foundation—the DexVerse physics engine, DexAgent general-purpose embodied agent, DexSense spatial perception terminal, and DexBrain intelligent algorithm brain—Kuawei Intelligence has built an integrated software and hardware capability system.
Currently, Kuawei Intelligence has deployed its solutions in over 50 application scenarios, iteratively accumulating more than 1,500 mature embodied models. These cover fields such as automotive parts, new energy, 3C electronics, aerospace, logistics, home appliances, chemicals, medical care, and education, serving specific scenarios including flexible sorting, flexible plug-in assembly, commercial retail, and cultural tourism services. Its clients include multiple leading companies in various industries.
Kuawei Intelligence has proposed the Physical Token economics concept, controlling costs and efficiency across the entire chain from data collection, model training, inference deployment, hardware body, to scenario operation and maintenance. Founder Professor Jia Kui stated that the company pursues the highest return on investment per Physical Token, ensuring that every set of data, every action, and every inference in the physical world can be efficiently converted into real commercial value. Kuawei Intelligence has validated its value loop in industrial, commercial, and elderly care/home scenarios, achieving revenue exceeding 100 million yuan.
An investor from Qianhai Ark said that Kuawei Intelligence relies on its self-developed simulation engine to efficiently generate synthetic data on a large scale, using world models to guide robots in scenarios such as retail, guided tours, and industrial applications. Its DexWorldModel achieved first place on the WorldArena Track2 leaderboard, proving the technical strength and deployment capability of the core team led by Professor Jia Kui. It is estimated that the company's total revenue this year could reach several hundred million yuan.
An investor from Lens Technology expressed optimism about the Sim2Real virtual-physical collaborative embodied intelligence track, noting that Kuawei Intelligence adheres to a pragmatic route of first learning physical laws in the virtual world before deploying in real industrial scenarios. Lens Technology will leverage its advantages in precision manufacturing and mass production of humanoid robot hardware, deeply collaborating with Kuawei Intelligence in simulation algorithms, perception software, and mass production of complete robots to promote industrial deployment.
